Neural networks
This chapter presents the use of ANNs to control the behavior of robots and presents an overview of learning in ANNs using the Hebbian rule.
Abstract
Chapter3 described reactive behaviors inspired by thework ofValentinoBraitenberg. The control of simple Braitenberg vehicles is very similar to the control of a living organism by its biological neural network. This term refers to the nervous system of a living organism, including its brain and the nerves that transmit signals through the body. Computerized models of neural networks are an active topic of research in artificial intelligence. Artificial neural networks (ANNs) enable complex behavior to be implemented using a large number of relatively simple abstract components that are modeled on neurons, the components of biological neural networks. This chapter presents the use of ANNs to control the behavior of robots. Following a brief overview of the biological nervous system in Sect. 13.1, Sect. 13.2 defines the ANN model and Sect. 13.3 shows how it can be used to implement the behavior of a Braitenberg vehicle. Section13.4 presents different network topologies. The most important characteristic of ANNs is their capability for learning which enables them to adapt their behavior. Section13.5 presents an overview of learning in ANNs using the Hebbian rule.